Marital Status,Marital Transitions,and Sleep Quality in Mid to Late Life

Abstract

This study aimed to understand how marital status and marital transitions were related to sleep quality in mid to late life and whether these findings differed by gender. Data from 2,872 participants 50–74 years old from the ORANJ BOWL^SM, a longitudinal panel study in New Jersey, were used. Marital status and sleep quality were examined in two waves that were approximately 10 years apart. Individuals in a significant romantic relationship and women had worse sleep quality than those in other marital status groups and men. Compared to individuals who remained married, individuals who remained divorced or widowed or who became widowed had better sleep quality, whereas those who became divorced had worse sleep quality; individuals who transitioned into marriage had better sleep quality than those who remained divorced or widowed. Marital status and gender appear important for at least some indices of sleep quality, an important predictor of late-life health.

Keywords

marital status gender sleep quality aging

Marital status differences in health have been well documented, with the dissolution of marriage being particularly detrimental for health (Williams & Umberson, 2004). Older adults who are widowed or divorced have been found to have poorer physical functioning, lower self-rated health, and greater mortality risk than their married counterparts (e.g., Brockmann & Klein, 2004; Pienta et al., 2000; Williams & Umberson, 2004). Health behaviors, such as sleep, have been identified as a potential mechanism whereby marital status affects health (Umberson & Montez, 2010). Sleep is an important behavior in the prevention and development of a number of chronic conditions in later life (Itani et al., 2017; Liu et al., 2013). One aspect of sleep commonly assessed in the literature, yet inconsistently defined, is sleep quality, which can refer a wide range of sleep experiences such as the latency to sleep onset, disruptive events during sleep, and an overall evaluation of restorative sleep (Krystal & Edinger, 2008). Regardless of how it is defined, poor sleep quality has been found to be a risk factor for poor health outcomes, including depression, anxiety, obesity, cardiovascular disease, and type 2 diabetes, in addition to cognitive impairments and overall mortality (Crowley, 2011; Zee & Turek, 2006). Sleep quality decreases with age, however, as a result of age-related changes in sleep architecture and circadian rhythm, as well as health issues and certain medications (Crowley, 2011; Roepke & Ancoli-Israel, 2010).

Marital status has been shown to account for variability in sleep quality (e.g., Kim et al., 2018; Whinnery et al., 2014), but little is known about this association in the later part of life in which sleep quality may have a more proximal impact on health. Moreover, few studies have examined how transitions into and out of marriage are related to changes in sleep quality. The current study therefore sought to understand how marital status was related to sleep quality in mid to late life and how marital transitions over an approximately 10-year period were related to sleep quality. Given evidence that men experience more detrimental consequences of marital dissolution than women (Williams & Umberson, 2004), this study also sought to examine whether associations between marital status, marital transitions, and sleep quality differed for men and women.

Marital status differences have been documented in both sleep duration and quality. In a number of studies using nationally representative samples of adults, being married has been found to be associated with more healthy sleep duration compared to being unmarried (e.g., Hale, 2005; Whinnery et al., 2014). Studies focused exclusively on middle-aged and older adults have reported similar results (Gu et al., 2010; Kim et al., 2018). For example, researchers in one study of older Chinese adults found that compared to being unmarried, being married was related to lower odds of very short or very long sleep duration (Gu et al., 2010). This study did not find a significant association between marriage and sleep quality, however. Yet some researchers have argued that sleep quality is just as consequential for health as sleep duration (Baker et al., 2009).

The research on marital status differences in sleep quality is more equivocal, however, potentially due to the gender and age ranges examined. Studies focused solely on women have found that married women report better overall sleep quality than unmarried women (Baker et al., 2009; Troxel et al., 2010). Yet in one study of midlife women (Troxel et al., 2010), the effects for sleep quality disappeared after controlling for sociodemographic and health characteristics, highlighting the need to include these variables in models that examine marital status and sleep quality. Findings from studies that have examined both genders are inconsistent, possibly due to the age group examined. For example, in one representative sample of British adults 16–74 years old, researchers found that divorced, separated, or widowed individuals were more likely to report sleep difficulties than their married or single counterparts (Arber et al., 2009). Whereas in another nationally representative sample of adults 18 years or older, researchers found that although married individuals had fewer problems remaining asleep compared to those who were divorced, they reported more disruptions (i.e., snoring) than those who were widowed, separated, or never married (Grandner et al., 2013). Finally, in a nationally representative sample of older adults, researchers did not find a difference in self-reported sleep quality between married and unmarried adults (Chen et al., 2015).

One theoretical framework that has been used to explain marital status differences in health behaviors more generally (Williams & Umberson, 2004), as well as sleep in particular (Meadows & Arber, 2015), is the “resource model,” which suggests that individuals confer benefits from marriage through greater resources. Specifically, having a bedpartner can increase feelings of safety and security that allow one to relax and fall asleep (Troxel et al., 2010). Another framework that can explain these differences is the “crisis model” (Williams & Umberson, 2004), which suggests that the dissolution of marriage causes stress, which in turn, interferes with sleep (Meadows & Arber, 2015). The extant literature on marital status differences in sleep has mainly focused on marital status at one point in time; there are limitations to this approach, however, as it does not allow for a test of whether transitioning into or out of marriage is more consequential for sleep. In one notable exception, Troxel and colleagues (2010) conducted a study of 370 midlife women in which they analyzed women’s marital histories 6–8 years prior to their study. Results revealed that women who were continually partnered (married or cohabitating) had better sleep quality than women who were unpartnered or who had lost or gained a partner during that time period.

Although the study by Troxel et al. (2010) advances an understanding of how marital transitions are related to sleep quality, men may have different health-related experiences than women during these transitions (Williams & Umberson, 2004). For example, some evidence suggests that compared to those who are married, insomnia symptoms and perceived sleep quality are worse for divorced/separated or widowed men than for women (Ohayon, 2002). Further, much of the research on marital status and sleep examines those who were married and cohabitating together (e.g., Baker et al., 2009; Chen et al., 2015), or those who are single and those in a non-marital romantic relationship together (e.g., Chapman et al., 2012). Middle-aged and older adults who are in significant, non-marital romantic relationships (that often involve cohabitation) may have different experiences than those who are married or single (Brown & Wright, 2017), so examining these adults separately from other marital status groups may yield different results. Moreover, examining the unmarried group as homogenous (e.g., Chen et al., 2015) may obscure important differences between individuals who are divorced, widowed, and never married. Finally, controlling for sociodemographic and health factors known to impact sleep quality is essential for understanding whether marital status has an effect beyond these known predictors.

The Current Study

Understanding how variations and changes in the social context may have implications for sleep quality is important for promoting healthy aging. The current study therefore aimed to elucidate how marital status and transitions into and out of marriage in mid to late life were related to sleep quality. Aim 1 examined baseline marital status differences in sleep quality and whether these differences varied by gender. Consistent with the resource model, it was expected that married individuals and those in significant romantic relationships would report better sleep quality than divorced/separated, widowed, or single individuals. This association was expected to be stronger for men than women. Aims 2 through 4 were focused on how marital continuity or transitions were related to sleep quality over time. Specifically, Aim 2 examined how individuals who remained divorced/separated, widowed, or single compared to those who remained married in sleep quality over time and whether these comparisons differed for men and women. Consistent with the resource model, it was expected that those who remained married would report better sleep quality than those who remained divorced/separated, widowed, or single and that these differences would be particularly pronounced for men. Aim 3 examined how individuals who transitioned out of marriage compared to those who remained married in sleep quality over the period of this transition, and whether these comparisons differed for men and women. Consistent with the crisis model, it was expected that those who transitioned out of marriage through divorce/separation or widowhood would report worse sleep quality than those who remained married, and that these differences would be particularly pronounced for men. Finally, Aim 4 examined how individuals who transitioned into marriage or a significant romantic relationship compared to individuals who remained unmarried over the period of this transition, and whether these comparisons differed for men and women. Consistent with the resource model, it was expected that those who transitioned into marriage or a significant romantic relationship would report better sleep quality than those who remained unmarried, and that these differences would be particularly pronounced for men.

Method

Procedure and Participants

Data from 2,872 individuals from the ORANJ BOWL (“Ongoing Research on Aging in New Jersey: Bettering Opportunities for Wellness in Life”) were used in the current study. The Rowan University Institutional Review Board approved study procedures. (The Rutgers University Institutional Review Board approved the use of data for the current analysis.) To recruit participants, cold calling and list-assisted random-digit-dialing procedures were used. Individuals were eligible to participate if they were 50–74 years old, resided in New Jersey, and could participate in a 1-hour telephone interview in English. The overall response rate for ORANJ BOWL was 58.7%; the cooperation rate was 72.9%. The ORANJ BOWL sample has been shown to be an adequate representation of the New Jersey population, and demographic characteristics of participants parallel those of the general U.S. population (Pruchno et al., 2010). More details about recruitment and other study procedures are provided elsewhere (Pruchno et al., 2010).

The initial sample included 5,688 participants who provided consent and participated in phone interviews at waves 1 and 6. Data were collected for wave 1 between November, 2006 and April, 2008 and for wave 6 between April, 2017 and May, 2019. Marital status was assessed at waves 1, 3, 5, and 6; sleep quality was only assessed at wave 1 and wave 6. Participants were excluded from the current analyses if marital status could not be classified due to missing data at wave 1 (n = 2), attrition at wave 6 (n = 2,650), or individuals experienced more than one marital transition or had inconsistent transitions between waves (n = 164). (See Heid et al., 2021 for details about attrition, retention, and re-engagement for this panel.) Compared to individuals who were excluded from analyses at baseline, those who were included were significantly younger, more educated, had higher incomes, less chronic health conditions, lower BMIs, more likely to be White, and less likely to have an emotional health issue (all ps ≤ .007). Participant characteristics by gender are reported in Table 1.

Table 1.

Participant Characteristics by Gender at Baseline (N = 2,872).

	Men (n = 1,042)	Women (n = 1,830)	Test of Gender Differences
Age, M(SD)	59.30(6.39)	60.04(6.82)	t(2870) = −2.91, p < .001
Marital Status, %			χ² (4) = 144.10, p < .001
Married^a	77.7%	57.5%
Significant romantic relationship	.8%	.9%
Divorced/separated^a	10.6%	19.9%
Widowed^a	4.0%	14.3%
Single	6.9%	7.3%
Race, %			χ² (4) = 29.31, p < .001
African American/Black^a	5.0%	8.6%
Asian/Pacific Islander	1.2%	1.2%
Native American^a	.4%	0%
White	90.6%	89.0%
Other^a	2.9%	1.2%
Education, %			χ² (4) = 48.77, p < .001
Less than high school	1.7%	1.9%
High school^a	19.1%	24.6%
Some college^a	19.2%	26.8%
Bachelor’s degree^a	27.7%	20.1%
At least some graduate education^a	32.2%	26.7%
Household income, %			χ² (2) = 88.25, p < .001
Less than $30,000^b	6.0%	13.3%
$30,000-$80,000^b	32.0%	41.9%
More than $80,000^b	62.0%	44.9%
Number of times married, M(SD)	1.20(.64)	1.16(.60)	t(2870) = 1.80, p = .07
Number of children in the household, M(SD)	.87(1.04)	.66(.91)	t(2870) = 5.11, p < .001
Body mass index, M(SD)	28.10(4.74)	28.24(6.64)	t(2870) = .65, p = .51
Number of physical health conditions, M(SD)	1.20(1.06)	1.75(1.31)	t(2870) = −12.30, p < .001
Presence of mental health issues, M(SD), %	14.5%	26.8%	χ² (1) = 57.79, p < .001

^a Statistically significant difference in adjusted standardized residuals determined using a Bonferroni correction of p < .005.

^bStatistically significant difference in adjusted standardized residuals determined using a Bonferroni correction of p < .008.

Measures

Marital status and marital transitions

To assess baseline marital status at wave 1, the following dummy variables were created from participants’ responses to questions about their previous and current marital status: a) married; b) in a significant (non-marital) romantic relationship; c) divorced/separated; d) widowed; and e) single (never married). To assess marital status continuity or transitions between waves 1 and 6, additional dummy variables were created (data from waves 3 and 5 were used, when available, to determine marital status continuity and transitions): a) continually married; b) continually divorced/separated; c) continually widowed; d) continually single; e) continually unmarried (aggregate of b–d); f) transition out of marriage into divorced/separated; g) transition out of marriage into widowhood; h) transition into marriage from unmarried (all groups); and i) transition into a significant romantic relationship from unmarried (all groups).

Sleep quality

To assess sleep quality at waves 1 and 6, four items were adapted from the Jenkins Sleep Scale (Jenkins et al., 1988) that have been used in studies assessing sleep quality in later life (e.g., Lee et al., 2017). Participants were asked how often they experienced the following three key indices of sleep quality on a 4-point scale (1 = “never,” 4 = “most of the time”), coded so that positive values indicated better sleep quality: sleep latency (one item: trouble falling asleep [reverse coded]); sleep disruptions (two items: waking up in the middle of the night [reverse coded]) and waking up early and being unable to fall back to sleep [reverse coded]); and restorative sleep (one item: feeling rested in the morning). Consistent with other research (e.g., Grandner et al., 2013) each of these items were examined separately, as using a global measure may obscure important differences in sleep experiences (Krystal & Edinger, 2008). In analyses that examined how marital status continuity or transitions were related to change in sleep quality over time (Aims 2–4), change scores were created for each sleep quality variable (sleep quality_wave6 – sleep quality_wave1).

Covariates

Gender was considered as a moderator in all analyses. Sociodemographic and health variables that have previously been examined in the literature on marital status and sleep were considered a priori as covariates. These included age, race (White vs. non-White), education, income, number of children in the household, body mass index (BMI; kg/m²), number of seven possible chronic physical health conditions (arthritis, hypertension, heart disease, cancer, diabetes, osteoporosis, lung disease), and presence of emotional health issues (“depression, anxiety, or any other emotional problems”).

Analytic Plan

SPSS version 27 was used for all analyses. First, data were checked for completeness. If data were missing on any of the marital status variables (at waves 1 or 6), participants were excluded from analyses. For most other variables, 0%–3% of data were missing. The only exception was number of children present in the household at wave 1 (22.2% missing data). As data were missing at random, the multiple imputation by chained equations procedure was performed in the wide data format to impute missing data (Young & Johnson, 2015).

To examine marital status differences in baseline levels of sleep quality and whether these differences varied by gender (Aim 1), a series of weighted least-squares regression analyses with dummy variables for each marital status were conducted. This method addresses unequal variances by incorporating information about the variance into the regression (Huitema, 2011). As each marital status group was used as a reference group to examine all possible comparisons, a Bonferroni correction was applied; significance was determine at α = .005. All regressions controlled for age, race, education, income, number of children in the household, BMI, number of chronic physical health conditions, and presence of mental health issues. Variables were entered in the following order: covariates (model 1); marital status and gender (model 2); and interactions between marital status × gender (model 3). If interactions were significant, models were examined within gender.

To examine how each marital status continuity or transition group compared to others on the change in sleep quality at wave 1 vs. wave 6, and whether these comparisons differed by gender (Aims 2–4), weighted fixed effect regression models using change scores (Allison, 2009) were conducted. Individuals continually in a significant romantic relationship were not used in these analyses due to very small sample sizes within each gender. Although all time-invariant variables (e.g., gender) are controlled for implicitly in fixed effects models, time-variant variables were included as covariates; these included change scores (wave 6 – wave 1) for age, number of children in the household, BMI, number of chronic physical health conditions, and presence of mental health issues. Variables were entered in the following order: all covariates (model 1); marital status (model 2); and interactions between marital status × gender (model 3). If interactions were significant, models were examined within gender.

Results

Frequencies of each marital status transition group and average changes in sleep quality between wave 1 and 6 by gender are presented in Table 2. In comparing marital transitions, men were significantly more likely than women to be continually married, whereas women were significantly more likely than men to be continually divorced/separated and widowed, and also more likely to transition into widowhood. In addition, among both men and women, there were significant differences between wave 1 and wave 6 in all sleep quality measures except feeling well-rested, with sleep quality being higher at wave 1 than at wave 6.

Table 2.

Marital Transitions and Change in Sleep Quality From Wave 1 to Wave 6 by Gender (N = 2,872).

	Men			Women
	(n = 1,042)			(n = 1,830)
Marital transition	χ² (8) = 237.05, p < .001
Continually married^a	755 (72.5%)			866 (47.3%)
Continually in a significant romantic relationship	5 (.4%)			7 (.4%)
Continually divorced/separated^a	74 (7.1%)			313 (17.1%)
Continually widowed^a	27 (2.6%)			246 (13.4%)
Continually single	68 (6.5%)			129 (7.0%)
Transition out of marriage to divorce/separation	24 (2.3%)			36 (2.0%)
Transition out of marriage to widowed^a	31 (3.0%)			150 (8.2%)
Transition into a significant romantic relationship	25 (2.5%)			34 (1.9%)
Transition into marriage	33 (3.2%)			49 (2.7%)
Sleep quality change	Wave 6 – Wave 1	t^b	p	Wave 6 – Wave 1	t^c	p
No trouble falling asleep	−.14	4.74	<.001	−.13	5.84	<.001
Not waking up in middle of the night	−.62	18.22	<.001	−.37	15.31	<.001
Not waking up early and being unable to fall back asleep	−.24	7.58	<.001	−.13	5.51	<.001
Feeling well-rested	−.04	1.48	.14	−.01	.35	.73

Note. Average time between wave 1 and wave 6 = 9.67 years.

^a Statistically significant difference in adjusted standardized residuals determined using a Bonferroni correction of p < .003.

^bTest of the difference between men’s sleep quality at wave 1 compared to wave 6; df = 1,041.

^cTest of the difference between women’s sleep quality at wave 1 compared to wave 6; df = 1,829.

Marital Status and Gender Differences in Sleep Quality at Baseline

Values for each sleep quality measure by marital status and gender at wave 1 are presented in Table 3. Weighted regression models tested differences in sleep quality by marital status, gender, and the interaction between marital status and gender (Aim 1).

Table 3.

Marital Status and Gender Differences in Sleep Quality at Baseline (N = 2,872).

	Marital Status and Gender
	Married (n = 1,897)		Significant Romantic Relationship (n = 61)		Divorced/ Separated (n = 535)		Widowed (n = 331)		Single (n = 212)		Model R²_adj
Sleep quality measure	Men	Women	Men	Women	Men	Women	Men	Women	Men	Women	Main effects	Interactions
No trouble falling asleep	2.97^e	2.66^ab	2.62^c	2.30^d	2.98^ce	2.70^ad	2.71^ce	2.68^abd	2.71^e	3.01^b	.08*	.09*
Not waking up in the middle of the night	2.77^a	2.55^b	2.48^a	2.08^b	2.61^a	2.45^b	2.60^a	2.54^b	2.55^a	2.58^b	.06*	.06
Not waking up early and being unable to fall back asleep	2.80^a	2.75^a	2.23^b	2.13^b	2.78^a	2.68^a	2.76^a	2.67^a	2.72^a	2.81^a	.03*	.03
Feeling well-rested	3.38^a	3.42^a	3.37^b	2.93^b	3.50^a	3.39^a	3.59^a	3.42^a	3.43^a	3.43^a	.09	.09

Note. Weighted least-squares regression analyses were used to compare groups by marital status, gender, and marital status × gender. Values in the same row that do not have any superscripts in common are significantly different from each other. Statistical significance was determined using a Bonferroni correction of p < .005. All regressions controlled for age, race, education, income, number of children in the household, BMI, number of chronic physical health conditions, and presence of mental health issues. Adjusted mean values are presented; higher values reflect better sleep quality.

* Significant change from the previous model (p < .05 or less).

A significant marital status difference was found for no trouble falling asleep, such that compared to individuals who were in a significant romantic relationship, those who were married (b = −.44, SE = .16, p = .005) and single (b = −.50, SE = .17, p = .004) reported significantly less trouble falling asleep. With single as the reference group, there was a significant interaction between married × gender (b = −.45, SE = .15, p = .002) and divorced × gender (b = −.55, SE = .18, p = .002) in predicting no trouble falling asleep. Within-gender analyses revealed that single women reported significantly less trouble falling asleep than divorced women (b = −.35, SE = .11, p = .002), but married women did not differ significantly from single women on this measure (b = −.21, SE = .10, p = .03). No significant differences between married and single men (b = .17, SE = .11, p = .14) nor divorced and single men (b = .24, SE = .14, p = .07) were found.

Significant marital status differences also were found for not waking up early and being unable to fall back asleep, such that compared to individuals in a significant romantic relationship, married (b = −.62, SE = .15, p < .001), divorced/separated (b = −.57, SE = .15, p < .001), widowed (b = −.53, SE = .16, p = .001), and single (b = −.63, SE = .16, p < .001) individuals reported waking up early and being unable to fall back asleep less. Likewise, compared to individuals in significant romantic relationships, those who were married (b = −.42, SE = .15, p = .005), divorced (b = −.42, SE = .15, p = .005), widowed (b = −.46, SE = .16, p = .003), and single (b = −.45, SE = .16, p = .004) reported feeling significantly more well-rested. These marital status differences did not differ by gender. No significant marital status differences were found for not waking up in the middle of the night.

Finally, significant gender differences in sleep quality were found for no trouble falling asleep (b = −.16, SE = .04, p < .001) and not waking up in the middle of the night (b = −.18, SE = .04, p < .001), with men reporting better sleep quality than women. No significant gender differences were found for not waking up early and being unable to fall back asleep and feeling well-rested.

Marital Status Continuity and Transitions and Sleep Quality Over Time

Results from fixed effects models that examined how individuals who remained divorced/separated, widowed, and single compared to individuals who remained married in change in sleep quality over time, and how these comparisons differed by gender (Aim 2) are presented in Table 4. Results revealed that compared to individuals who remained married, those who remained divorced and widowed reported waking up in the middle of the night less; those who remained widowed also reported waking up early and being unable to fall back asleep less. There was also a significant interaction with gender in comparing individuals who remained widowed and those who remained married in predicting not waking up in the middle of the night, but within gender analyses did not find a difference between men and women (i.e., the main effects for men and women were both significant in the same direction). In addition, there was a significant interaction with gender in comparing individuals who remained divorced/separated and those who remained married in predicting feeling well-rested. Within-gender analyses revealed a significant effect for women (b = .26, SE = .03, p = .001), but not men (b = −.17, SE = .16, p = .29), such that women who remained divorced/separated reported feeling more well-rested than women who remained married. No significant differences emerged between those who remained married and those who remained divorced/separated, widowed, or single in no trouble falling asleep.

Table 4.

Fixed Effects Regression Results for Marital Status Continuity Differences in Sleep Quality From Wave 1 to Wave 6 (N = 2,478).

	No Trouble Falling Asleep		Not Waking up in the Middle of the Night		Not Waking up Early and Being Unable to Fall Back Asleep		Feeling Well-Rested
	B	SE	B	SE	B	SE	B	SE
Main effects
Continually divorced/separated	.13	.36	.13**	.04	<.01	.04	.24***	.06
Continually widowed	.47	.43	.19***	.05	.19***	.05	.07	.08
Continually single	.69	.62	.03	.06	.06	.05	.16	.09
	R²_adj = .38*		R²_adj = .03*		R²_adj = .43*		R²_adj = .01*
Interactions
Continually divorced/separated × gender	−.15	.68	.01	.10	.07	.09	.46**	.17
Continually widowed × gender	−.25	.92	.20*	.17	−.06	.14	.47	.30
Continually single × gender	−.60	.89	.02	.13	−.05	.11	.21	.18
	R²_adj = .40		R²_adj = .03		R²_adj = .43		R²_adj = .01

Note. Reference group = continually married. All models presented controlled for change scores (wave 6 – wave 1) in age, number of children in the household, BMI, number of chronic physical health conditions, and presence of mental health issues. SE = standard error. Significance for R² reflects significant change from the previous model (p < .05 or less).

*p < .05. **p < .01. ***p < .001.

Results from fixed effects models that examined how individuals who transitioned out of marriage into divorce/separation or widowhood compared to individuals who remained married in sleep quality over the period of the transition, and how these comparisons differed by gender (Aim 3) are presented at the top of Table 5. Results revealed that compared to individuals who got divorced/separated, those who remained married reported waking up in the middle of the night less. There was a significant interaction with gender in comparing individuals who transitioned to widowhood and those who remained married in predicting not waking up in the middle of the night. Women who became widowed reported waking up in the middle of the night less than women who remained married (b = .45, SE = .20, p = .04), but this effect was not significant for men (b = −.06, SE = .15, p = .69). Finally, divorced individuals reported feeling more well-rested than married individuals. No significant differences emerged between those who remained married and those who got divorced/separated or widowed in no trouble falling asleep or not waking up early and being unable to fall back asleep.

Table 5.

Fixed Effects Regression Results for Transition Out of and Into Marriage or a Significant Romantic Relationship and Change in Sleep Quality From Wave 1 to Wave 6.

	No Trouble Falling Asleep		Not Waking up in the Middle of the Night		Not Waking up Early and Being Unable to Fall Back Asleep		Feeling Well-Rested
Transition out of marriage (N = 1,872)^a	B	SE	B	SE	B	SE	B	SE
Main effects
Transition into divorce/separated	−.02	.13	−.26**	.08	−.09	.22	.31*	.14
Transition into widowed	.32	.24	.28***	.05	.20	.37	<.01	.10
	R²_adj = .34*		R²_adj = .05*		R²_adj = .12*		R²_adj = < .01
Interaction effects
Transition into divorce/separated × gender	.19	.52	.27	.40	−.23	.51	.48	.29
Transition into widowed × gender	.73	.57	.49*	.21	.56	.48	.22	.26
Transition into a relationship (N = 998)^b	R²_adj = .34		R²_adj = .05		R²_adj = .12		R²_adj = < .01
Main effects
Transition into marriage	−.11	.14	−.13	.12	−.11	.41	−.04	.15
Transition into significant romantic relationship	.39	.24	−.01	.12	.50	.34	.20	.19
	R²_adj = .17*		R²_adj = .01		R²_adj = .45*		R²_adj = .11
Interaction effects
Transition into marriage × gender	.34	.31	.20	.27	−.05	1.01	.66*	.32
Transition into significant romantic relationship × gender	−.69	.37	.53	.37	.84	.95	−.12	.37
	R²_adj = .19*		R²_adj = .01		R²_adj = .45		R²_adj = .11

Note. All models presented controlled for change scores (wave 6 – wave 1) in age, number of children in the household, BMI, number of chronic physical health conditions, and presence of mental health issues. SE = standard error. Significance for R² reflects significant change from the previous model (p < .05 or less).

^a Reference group = continually married; ^bReference group = continually unmarried.

*p < .05. **p < .01. ***p < .001.

Finally, results from fixed effects models that examined how individuals who transitioned into marriage or a significant romantic relationship compared to individuals who remained unmarried in sleep quality over the period of the transition, and how these comparisons differed by gender (Aim 4) are presented at the bottom of Table 5. The only significant finding was an interaction with gender in comparing individuals who transitioned into marriage compared to those who remained unmarried in predicting feeling-well-rested, but within gender analyses did not find significant main effects for men or women. No other significant differences emerged when comparing individuals who remained unmarried and those who transitioned into marriage or a significant romantic relationship in any of the other sleep quality measures. Post-hoc analyses were undertaken to determine whether the transition into marriage or a romantic relationship was associated with a change in sleep quality when compared to a specific unmarried group (continually divorced/separated, widowed, single). Two significant findings emerged: 1) individuals who remained divorced reported feeling more well-rested than those who transitioned into marriage (B = −.69, SE = .26, p = .01) and 2) individuals who remained widowed reported less difficulty falling asleep than those who transitioned into marriage (B = −.25, SE = .10, p = .02).

Discussion

Despite the critical role sleep plays in the prevention and management of chronic health conditions, as well as overall cognitive functioning, in mid to late life (Itani et al., 2017; Liu et al., 2013), few studies have examined the link between marital status—particularly marital transitions—and sleep among both older men and women. The current study used the resource model and crisis model (Williams & Umberson, 2004) as frameworks for understanding these associations.

Marital Status Differences in Sleep Quality

Contrary to expectations and the resource model, results revealed that, compared to individuals in all other marital status groups, those in a significant romantic relationship experienced worse quality sleep on three of the four items. In this sample, individuals in significant romantic relationships were together significantly less time than those who were married (data not shown); thus, individuals in these relationships may be adjusting to the presence of a bed partner (Troxel et al., 2010), which research suggests may have a negative impact on sleep (Pankurst & Horne, 1994). In addition, sexual activity may be more frequent in these types of relationships than others (DeLamater & Sill, 2005), further interfering with sleep. Finally, individuals in a significant romantic relationship may not have consistent sleep environments, as “living apart together” arrangements are becoming more common in later life (Benson & Coleman, 2016). Although these results provide preliminary evidence about the sleep quality of individuals in significant romantic relationships, they should be interpreted with caution, given the relatively small number of participants who were in this category. Future research with larger samples of individuals in this type of relationship would increase confidence in these findings.

In line with expectations and the crisis model, results also revealed that divorced women had significantly more trouble falling asleep than single women. This finding is consistent with the broader literature that divorced women have high rates of sleep disturbances (Hale, 2005). Lower sleep quality among divorced women can occur “by negatively influencing mood and affect and leading to endocrine and autonomic dysregulation” (Troxel et al., 2010, p. 974). Contrary to expectations and the resource model, no significant differences in sleep quality emerged between individuals who were married and those who were in any of the unmarried groups. Although this finding was unexpected, it is consistent with at least one other study of older adults in which researchers also did not find differences in self-reported sleep quality between married and unmarried adults (Chen et al., 2015). Other studies that have found marital status differences in sleep quality were mainly focused on a broader range of age groups (e.g., Arber et al., 2009) or only women (e.g., Troxel et al., 2010). Another possible reason for the lack of marital status differences in sleep quality is that the institution of marriage may not be as important for sleep quality as the underlying relationship processes associated with marriage. For example, there is evidence that in poorer quality marriages, sleep quality is also poor (e.g., Lee et al., 2017). Thus, although results from the current study and the study by Chen and colleagues (2015) suggest that simply being married may not confer benefits with regard to subjective perceptions of sleep quality, more research is needed in later life to establish the validity of these conclusions.

Marital Status Continuity and Transitions and Sleep Quality

Although marital status differences provide a “snapshot” of how middle-aged and older adults in different relationship statuses compare to each other on measures of sleep quality, they do not allow for a test of whether it is being married (or partnered) that is beneficial to sleep or whether it is the dissolution of marriage (or the absence of a partner) that is detrimental to sleep. When examining marital status continuity and transitions and corresponding changes in sleep quality over time, a different pattern than marital status differences in sleep quality emerged. Contrary to expectations and the crisis model, women who remained divorced/separated and both men and women who remained widowed reported better sleep quality than those who remained married on three of the sleep quality measures—waking up in the middle of the night, waking up early and being unable to fall back asleep, and feeling well-rested. One possible explanation for these findings is that sleeping with a spouse who develops health issues (e.g., sleep apnea) could interfere with an individual’s own sleep (Richter et al., 2016). In addition, as many older spouses, particularly women, are caregivers (AARP & National Alliance on Caregiving [NAC], 2020), those who remain married are more likely to take on caregiving responsibilities in later adulthood. Caregiving, when accompanied by negative affect, has been found to be related to poorer sleep quality (Brummett et al., 2006). Yet there were no significant differences in sleep quality between those who were continually married and those who were continually single. As most individuals in the single category were never married, it is possible that sleep quality is only better among those who have a reference point of having previously had a spouse (i.e., those who were divorced/separated and widowed) versus never having a spouse in the first place.

Consistent with the crisis model (Williams & Umberson, 2004) and in partial support of hypotheses, those who became divorced over the approximately 10 year period had a corresponding decrease in one aspect of sleep quality—specifically, waking up in the middle of the night—compared to those who remained married. The dissolution of marriage due to divorce in later life may result in stressors such as financial difficulties and reduced contact with older children (Brown & Wright, 2017), which, as noted above, could lead to sleep disruptions through neuroendocrine and autonomic pathways (Troxel et al., 2010). Future research would benefit from understanding what aspects of the transition into divorce influence sleep quality to confirm this conjecture. Interestingly (and contrary to expectations and the crisis model), individuals who became divorced also reported feeling more well-rested than those who remained married. The different items used to assess sleep quality could provide insight into to these seemingly contradictory findings. Although all measures examined in this study were subjective, “feeling well-rested” is arguably more subjective than the other measures of sleep quality, given individual differences in the need for a certain amount of sleep (Van Dongen et al., 2005). Thus, even if divorced individuals felt well-rested, their reports of waking up in the middle night suggest their sleep was disturbed to some extent.

In contrast to findings for individuals who became divorced, individuals who became widowed reported waking up early and being unable to fall back asleep and waking up in the middle of the night (among women only) less than individuals who remained married. Unlike precursors to divorce in late life (Brown & Wright, 2017), spouses’ ill health may be a precursor to widowhood; widowed individuals (especially women) thus may be relieved of caring for an ill spouse (AARP & NAC, 2020), which in turn, has a positive impact on sleep quality. Another study (e.g., Reynolds et al., 1992) found an increased risk of sleep disturbances among widowed older adults, but only among those who were depressed. Affective reactions to spousal bereavement therefore may be more important than bereavement itself in influencing sleep quality; as the current study examined changes in sleep quality over an approximately 10 year period, it is possible the individuals in this study who became widowed already adjusted psychologically to the loss of their spouse (Sasson & Umberson, 2014) and thus reported improvement in their sleep quality. Findings may differ for individuals who are recently widowed and have adverse psychological reactions to their bereavement.

Contrary to expectations and the resource model, there were no significant effects for individuals transitioning into marriage or a significant romantic relationship when compared to the continually unmarried as an overall group. Post-hoc analyses that examined each continually unmarried group separately, however, uncovered a different pattern of results. Specifically, those who were continually divorced and widowed reported better sleep quality than those who transitioned into marriage on two aspects of sleep quality (feeling well-rested and no trouble falling asleep, respectively). These findings were consistent with other studies (e.g., Troxel et al., 2010) that gaining a bedpartner through marriage negatively affects sleep. The null findings for individuals who transitioned into a significant romantic relationship could reflect the possibility that not all of these individual who transitioned into this status were sharing a bed (at least not yet). These post-hoc findings highlight the importance of examining unmarried individuals in distinct groups.

Finally, no significant findings were found for marital transition differences in one measure of sleep quality, namely, waking up early and being unable to fall back asleep. Notably, none of the covariates were significantly associated with a change in this measure of sleep quality either. It is possible that other factors not assessed in the current study (e.g., sleep environment; Ancoli-Israel et al., 2008) are more important predictors of waking up early and being unable to fall back asleep. Alternatively, it may be that intrinsic factors associated with aging, such as a shift in circadian rhythms (Roepke & Ancoli-Israel, 2010) are mainly responsible for changes in this measure of sleep quality with age, as evidenced in the current study.

Gender Differences in Sleep Quality

Although most marital status and marital transition differences in sleep quality did not differ by gender, consistent with the broader literature, gender differences in sleep quality emerged, with women reporting worse quality sleep than men on two of the sleep quality measures: trouble falling asleep and waking up in the middle of the night. Compared to older men, older women have more sleep-related disorders (e.g., restless legs syndrome, insomnia), mental health issues, caregiving roles, and hormonal changes during and after menopause that interfere with sleep (Guidozzi, 2015; Mallampalli & Carter, 2014). Moreover, women are more likely to report poorer sleep quality than men, even when polysomnographic evidence does not support these differences (Mallampalli & Carter, 2014).

Limitations and Future Directions

The results of this study should be interpreted in the context of limitations. First, sleep quality was only assessed at two waves, thus precluding an examination of how the timing of marital status changes across multiple waves were associated with corresponding changes in sleep quality across those waves. For example, it is possible that marital transitions have a stronger effect on sleep quality in the earlier stages of these transitions, but after an adjustment period in a new marital status, sleep may be less affected. Further, sleep quality was subjectively measured via self-report using a brief measure. Some researchers have argued that subjective perceptions of sleep are still important because they can affect overall quality of life, motivate people to seek medical help, and are associated with adverse health outcomes and mortality (Baker et al., 2009). In addition, there was no assessment of whether participants who were married or in romantic relationships were actually sleeping with their partner, nor was there an assessment of relationship quality, which may have a different association with sleep quality than relationship status (Chen et al., 2015). Finally, the sample was mostly White and of relatively high socioeconomic status (SES), and a number of participants had to be excluded due to missing data or attrition (who differed significantly from those included in analyses), thus limiting generalizability to racial/ethnic minorities and individuals of lower SES. Future studies that assess sleep quality at multiple time points, use more objective measures, and include a more diverse sample (including a greater number of individuals who are in non-marital romantic relationships) would contribute to a more comprehensive understanding of how sleep may be impacted by marital status. Future research also should focus on exploring how long sleep quality is impacted by marital transitions as well as the mechanisms and conditions in which these transitions are related to sleep quality.

Conclusion

Despite these limitations, this study contributes to a greater understanding of how sleep quality may differ as a function of marital status, marital transitions, and gender in mid to late life. In general, transitions into and out of marriage were more detrimental for sleep quality than marital status continuity. In addition, instead of the predicted interactive effects between marital status and gender, additive effects of marital status and gender on sleep quality were more common. Importantly, results were robust after controlling for sociodemographic and health factors known to impact sleep.

The findings from this study extend past research by highlighting the importance of examining indices of sleep quality separately, as treatment efforts may differ depending on the sleep impairment older adults are experiencing (Krystal & Edinger, 2008). Further, “unpacking” the unmarried group, including those in a significant (non-marital) romantic relationship, allows for a better understanding of which groups of middle-aged and older adults are most vulnerable to different sleep impairments. As difficulty falling and staying asleep, as well as non-restorative sleep, increase older adults’ risk for falls, impaired concentration and memory, mental and physical health conditions, and decreased quality of life (Ancoli-Israel et al., 2008; Crowley, 2011), providers can tailor their discussions and treatments with patients most at risk for these different indices of poor sleep quality.

Footnotes

Author’s Note

Data for this study were obtained upon permission from Rachel Pruchno. ORANJ BOWL^SM (Ongoing Research on Aging in New Jersey: Bettering Opportunities for Wellness in Life) Project, Stratford, NJ: Rowan University School of Osteopathic Medicine.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: ORANJ BOWL^SM was supported by the University of Medicine and Dentistry of New Jersey (waves 1 and 2); the Assistant Secretary for Preparedness & Response (1 HITEP130008; wave 4); and the National Institute on Aging (R01AG046463; waves 5 and 6).

ORCID iD

Kristin J. August

Author Biography

Kristin J. August, PhD. is an Associate Professor of Psychology and core faculty member of the Health Sciences Center at Rutgers University, Camden.

References

AARP and National Alliance for Caregiving. Caregiving in the United States. (2020, May). AARP. https://doi.org/10.26419/ppi.00103.001

Allison

. (2009). Fixed effects regression models. Sage.

Ancoli-Israel

Ayalon

Salzman

(2008). Sleep in the elderly: Normal variations and common sleep disorders. Harvard Review of Psychiatry, 16(5), 279–286. https://doi.org/10.1080/10673220802432210

Arber

Bote

Meadows

(2009). Gender and socio-economic patterning of self-reported sleep problems in Britain. Social Science & Medicine, 68, 281–289. https://doi.org/10.1016/j.socscimed.2008.10.016

Baker

F. C.

Wolfson

A. R.

Lee

K. A.

(2009). Association of sociodemographic, lifestyle, and health factors with sleep quality and daytime sleepiness in women: Findings from the 2007 national sleep foundation “sleep in America poll”. Journal of Women’s Health, 18(6), 841–849. https://doi.org/10.1089/jwh.2008.0986

Benson

J. J.

Coleman

(2016). Older adult descriptions of living apart together. Family Relations, 65, 439–449. https://doi.org/10.1111/fare.12203

Brockmann

Klein

(2004). Love and death in Germany: The marital biography and its effect on mortality. Journal of Marriage and Family, 66(3), 567–581. https://doi.org/10.1111/j.0022-2445.2004.00038.x

Brown

S. L.

Wright

M. R.

(2017). Marriage, cohabitation, and divorce in later life. Innovation in Aging, 1(2), https://doi.org/10.1093/geroni/igx015

Brummett

B. H.

Babyak

M. A.

Siegler

I. C.

Vitaliano

P. P.

Ballard

E. L.

Gwyther

L. P.

Williams

R. B.

(2006). Associations among perceptions of social support, negative affect, and quality of sleep in caregivers and noncaregivers. Health Psychology, 25(2), 220. https://doi.org/10.1037/0278-6133.25.2.220

10.

Chapman

D. P.

Wheaton

A. G.

Perry

G. S.

Sturgis

S. L.

Strine

T. W.

Croft

J. B.

(2012). Household demographics and perceived insufficient sleep among US adults. Journal of Community Health, 37, 344–349. https://doi.org/10.1007/s10900-011-9451-x

11.

Chen

J. H.

Waite

L. J.

Lauderdale

D. S.

(2015). Marriage, relationship quality, and sleep among US older adults. Journal of Health and Social Behavior, 56(3), 356–377. https://doi.org/10.1177/0022146515594631

12.

Crowley

(2011). Sleep and sleep disorders in older adults. Neuropsychology Review, 21(1), 41–53. https://doi.org/10.1007/s11065-010-9154-6

13.

DeLamater

Sill

(2005). Sexual desire in later life. Journal of Sex Research, 42(2), 138–149. https://doi.org/10.1080/00224490509552267

14.

Grandner

M. A.

Petrov

M. E. R.

Rattanaumpawan

Jackson

Platt

Patel

N. P.

(2013). Sleep symptoms, race/ethnicity, and socioeconomic position. Journal of Clinical Sleep Medicine, 9(9), 897–905. https://dx-doi-org.web.bisu.edu.cn/10.5664/jcsm.2990

15.

Sautter

Pipkin

Zeng

(2010). Sociodemographic and health correlates of sleep quality and duration among very old Chinese. Sleep, 33, 601–610. https://doi.org/10.1093/sleep/33.5.601

16.

Guidozzi

(2015). Gender differences in sleep in older men and women. Climacteric, 18(5), 715–721.

17.

Hale

(2005). Who has time to sleep? Journal of Public Health, 27(2), 205–211.

18.

Heid

A. R.

Cartwright

F. P.

Wilson-Genderson

Pruchno

(2021). Understanding attrition and bolstering retention in a longitudinal panel of older adults: ORANJ BOWL. Innovation in Aging, 5, 1–13.

19.

Huitema

B. E.

(2011). The analysis of covariance and alternatives: Statistical methods for experiments, quasi-experiments, and single-case studies. John Wiley & Sons.

20.

Itani

Jike

Watanabe

Kaneita

(2017). Short sleep duration and health outcomes: A systematic review, meta-analysis, and meta-regression. Sleep Medicine, 32, 246–256. https://doi.org/10.1016/j.sleep.2016.08.006

21.

Jenkins

C. D.

Stanton

B. A.

Niemcryk

S. J.

Rose

R. M.

(1988). A scale for the estimation of sleep problems in clinical research. Journal of Clinical Epidemiology, 41(4), 313–321.

22.

Kim

Lee

J. A.

Park

H. S.

(2018). Health behaviors and illness according to marital status in middle-aged Koreans. Journal of Public Health, 40(2), e99–e106. https://doi.org/10.1093/pubmed/fdx071

23.

Krystal

A. D.

Edinger

J. D.

(2008). Measuring sleep quality. Sleep Medicine, 9, S10–S17.

24.

Lee

J. H.

Chopik

W. J.

Schiamberg

L. B.

(2017). Longitudinal associations between marital quality and sleep quality in older adulthood. Journal of Behavioral Medicine, 40(5), 821–831. https://doi.org/10.1007/s10865-017-9850-2

25.

Liu

Wheaton

A. G.

Chapman

D. P.

Croft

J. B.

(2013). Sleep duration and chronic diseases among US adults age 45 years and older: Evidence from the 2010 behavioral risk factor surveillance system. Sleep, 36(10), 1421–1427. https://doi.org/10.5665/sleep.3028

26.

Mallampalli

M. P.

Carter

C. L.

(2014). Exploring sex and gender differences in sleep health: A society for women’s health research report. Journal of Women’s Health, 23(7), 553–562. https://doi.org/10.1089/jwh.2014.4816

27.

Meadows

Arber

(2015). Marital status, relationship distress, and self-rated health: What role for “sleep problems”? Journal of Health and Social Behavior, 56(3), 341–355.

28.

Ohayon

M. M.

(2002). Epidemiology of insomnia: What we know and what we still need to learn. Sleep Medicine Reviews, 6(2), 97–111. https://doi.org/10.1053/smrv.2002.0186

29.

Pienta

A. M.

Hayward

M. D.

Jenkins

K. R.

(2000). Health consequences of marriage for the retirement years. Journal of Family Issues, 21(5), 559–586.

30.

Pruchno

R. A.

Wilson-Genderson

Rose

Cartwright

(2010). Successful aging: Early influences and contemporary characteristics. The Gerontologist, 50(6), 821–833.

31.

Reynolds, III

C. F.

Hoch

C. C.

Buysse

D. J.

Houck

P. R.

Schlernitzauer

Frank

Mazumbar

Kupfer

D. J.

(1992). Electroencephalographic sleep in spousal bereavement and bereavement-related depression of late life. Biological Psychiatry, 31(1), 69–82.

32.

Richter

Adam

Geiss

Peter

Niklewski

(2016). Two in a bed: The influence of couple sleeping and chronotypes on relationship and sleep. An overview. Chronobiology International, 33(10), 1464–1472. https://doi.org/10.1080/07420528.2016.1220388

33.

Roepke

S. K.

Ancoli-Israel

(2010). Sleep disorders in the elderly. Indian Journal of Medical Research, 131(2), 302–310.

34.

Sasson

Umberson

D. J.

(2014). Widowhood and depression: New light on gender differences, selection, and psychological adjustment. Journals of Gerontology. Series B: Psychological Sciences and Social Sciences, 69(1), 135–145.

35.

Troxel

W. M.

Buysse

D. J.

Matthews

K. A.

Kravitz

H. M.

Bromberger

J. T.

Sowers

Hall

M. H.

(2010). Marital/cohabitation status and history in relation to sleep in midlife women. Sleep, 33(7), 973–981. https://doi.org/10.1093/sleep/33.7.973

36.

Umberson

Montez

J. K.

(2010). Social relationships and health: A flashpoint for health policy. Journal of Health and Social Behavior, 51(Suppl), S54–S66.

37.

Van Dongen

H. P.

Vitellaro

K. M.

Dinges

D. F.

(2005). Individual differences in adult human sleep and wakefulness: Leitmotif for a research agenda. Sleep, 28(4), 479–498.

38.

Whinnery

Jackson

Rattanaumpawan

Grandner

M. A.

(2014). Short and long sleep duration associated with race/ethnicity, sociodemographics, and socioeconomic position. Sleep, 37(3), 601–611. https://doi.org/10.5665/sleep.3508

39.

Williams

Umberson

(2004). Marital status, marital transitions, and health: A gendered life course perspective. Journal of Health and Social Behavior, 45(1), 81–98.

40.

Young

Johnson

D. R.

(2015). Handling missing values in longitudinal panel data with multiple imputation. Journal of Marriage and the Family, 77(1), 277–294.

41.

Zee

P. C.

Turek

F. W.

(2006). Sleep and health: Everywhere and in both directions. Archives of Internal Medicine, 166(16), 1686–1688. https://doi.org/10.1001/archinte.166.16.1686